OpenAI’s Trust Problem Isn’t Just About the Tech

OpenAI dropped a set of policy recommendations this week about how to handle superintelligence — you know, the hypothetical moment when AI outperforms the smartest humans even with assistance. The document is full of noble language: “keep people first,” “clear-eyed” risk monitoring, warnings about AI evading human control or governments using it to undermine democracy. If we don’t get this right, OpenAI says, people will be harmed.

Sounds great on paper.

But on the exact same day, The New Yorker published a massive investigation into whether CEO Sam Altman can actually be trusted to deliver on those promises. Reading them side by side is like watching someone swear a sacred oath while their coworkers whisper that they’ve been caught lying before.

The timing couldn’t be worse — or more revealing. OpenAI wants us to believe it’s the responsible steward of humanity’s future. Meanwhile, insiders are telling journalists they don’t trust the guy running the show.

This isn’t some fringe criticism from random ex-employees either. The New Yorker piece goes deep into Altman’s track record, his management style, and the growing gap between OpenAI’s public messaging and internal reality. I’ve been around long enough to know that when multiple insiders speak on the record about trust issues at the CEO level, something is rotten.

Let’s be real: OpenAI has been playing both sides for a while. They warn about AI risks in press releases while racing to deploy products. They talk about transparency while keeping key safety details locked up. They position themselves as the responsible alternative to Google and Meta, but their internal culture sounds increasingly chaotic.

The policy document itself isn’t bad. It’s actually thoughtful about monitoring for extreme scenarios and advocating for broad benefit distribution. But documents don’t build trust — people do. And when the person at the top has a credibility problem with his own staff, those words start looking like decorations.

I’m not saying Altman is a villain. He’s clearly smart and ambitious. But OpenAI’s biggest challenge right now might not be technical alignment or safety research. It might be convincing people — including their own employees — that they mean what they say.

Because if you can’t trust the CEO, why would anyone trust the company’s vision for the future?